DAGA 2014 Oldenburg

Acoustic in-ice positioning in the Enceladus Explorer project

Franziska Scholz, Dmitry Eliseev, Dirk Heinen, Sebastian Verfers, Christopher Wiebusch, Simon Zierke III. Physikalisches Institut, RWTH Aachen University, D-52056 Aachen, Germany Email: [email protected]

Introduction One of the most challenging tasks for humanity is the search for extraterrestrial life. A promising candidate for this search is Enceladus, a Saturn moon, which is com- pletely covered by an ice crust. Recently, the Cassini spacecraft found evidence for a large underground ocean of liquid water at the south pole of Enceladus [1]. In earlier measurements Cassini found frozen water parti- Figure 1: Sketch of the APS measurement technique: Emit- cles, salt and organic molecules in geysir-like jets ejected ters form a net on the ice surface. Receivers are situated in near the south pole [2]. The combination of both obser- the head of the probe and measure the arrival time of the vations strengthens the possibility of the existence of ex- emitted signal [3]. traterrestrial life on Enceladus. For further investigation it becomes essential to probe the liquid water directly. The Enceladus Explorer Project has been funded by the German Space Agency DLR as a first step toward the accomplishment of such a mission.

Enceladus Explorer project The goal of the Enceladus Explorer project is to develop a maneuverable probe for a terrestrial test scenario where samples are to be taken from a subglacial water pocket at Figure 2: Sensor design: The piezo element with the current the Blood Falls in Antarctica in November 2014 [3]. The front end electronics is situated in the APS sensor hull (left). used probe is called “IceMole“. Within the Enceladus The sensor hulls are integrated in the IceMole head [3] (right). Explorer project it has been successfully tested during two field tests, one on the Morteratsch Glacier in Switzer- land in June 2013 and one on the Canada Glacier in tion. A sketch of the APS method is shown in figure 1. Antarctica in November 2013. Prior to the final test there The ultrasound emitters are commercial sonar transduc- will be another field test in June 2014 on the Morteratsch ers which emit sinus bursts with a frequency of 18 kHz Glacier in Switzerland. and a length of 11 periods [5]. Each ultrasound receiver [6] consists of a piezo element whose signals are directly IceMole processed by the APS front end electronics [7]. The cur- The IceMole is a melting probe with an ice screw for rent front end electronics filters out frequencies below forward thrust [4]. Partial heating of the IceMole head 10 kHz and above 300 kHz and therefore has a rather allows to drive curved trajectories through the ice. To broad passband [6]. The sensor design is shown in figure monitor and control these trajectories a navigation sys- 2. The desired detection range of the system is at least tem has been developed. This navigation system consists 100 m [3]. of an inertial measurement unit and a magnetometer as well as an acoustic navigation system containing an ultra- Optimization of the signal detection sonic reconnaissance system, which explores the fore-field of the probe and an acoustic positioning system, which The main challenge in analyzing the raw data is to find a determines the absolute position of the IceMole [3]. robust method for a precise determination of the arrival time of the signal. The five sigma method (FSM) searches Acoustic Postitioning System for the first time where the noise level is surpassed by a factor of five. The noise level is defined as the stan- The Acoustic Positioning System (APS) uses six ultra- dard deviation of the first millisecond of data, which is sound emitters embedded in the upper ice layers as well recorded before a signal is being transmitted. The APS as four ultrasound receivers in the IceMole head. The reduces the noise by averaging over 32 waveforms and position of the IceMole is determined by measuring the thus reducing miss-identifications. However if the noise arrival times of the emitted pulses, converting them into level is too high or if there are other large signals recorded distances and using trilateration to calculate the posi- beside the emitted one the FSM fails. A waveform where

692 DAGA 2014 Oldenburg

signal at about 3 ms can be found by the FSM and the transient signals at the end of the waveform were able to be significantly reduced in comparison to the emitted signal. To compare the noise reduction of different fil- ters the best choice is to look at the signal-to-noise ratio (SNR). The SNR of the Fourier filtered data is about 7.8 while for the IIR filtered data the SNR is about 17.7 which is a factor of 2.3 higher. The investigation of the optimal filter characteristics was done as a design study for the development of new sensor electronics for APS. The results strongly suggest the use of analog filters in future APS sensor electronics.

Summary & Outlook In this contribution a design study for the use of analog filters in the APS sensor electronics was presented. The differences between the compared filters were pointed out on the basis of data, which was taken during the field test in Antarctica. It was shown that the use of analog filters in the APS sensor electronics is absolutely recom- mended. Based on the design studies, improved front Figure 3: (1) Measured data averaged over 32 waveforms, (2) Fourier spectrum of the data, frequency response of the end electronics with an optimized two stage analog filter Fourier filter (red) and the IIR filter (green), (3) Fourier fil- were developed and are in production. First test mea- tered data, (4) IIR filtered data. The dashed horizontal lines surements with a prototype in the laboratory indicate an indicate five times the standard deviation of the noise. The improvement compared to the old front end electronics. vertical lines mark the determined arrival time. The different Further test measurements will be done during the field plots will be explained more precisely in the chapter on the test in Switzerland in June 2014 to confirm this improve- optimization of the signal detection. ment under realistic conditions.

References such problems occur is shown in figure 3.1. The data was [1] L. Iess et.al: The Gravity Field and Interior Structure taken during the field test in Antarctica with a distance of Enceladus. Science, vol. 344 no. 6179, pp. 78-80, between emitter and receiver of about 7 m. Therefore one 2014 expects to detect a signal at around 3 ms, including 1 ms of noise. It is obvious that no signal exceeds the noise [2] C. C. Porco et.al: Cassini Observes the Active South level at around 3 ms. Instead there are two other signals Pole of Enceladus. Science, vol. 311 no. 5766, pp. at the end of the waveform, which are detected by the 1393-1401, 2006 FSM. Closer investigations using a time resolved Fourier [3] G. Ameres et.al: Design Freeze Review Field Test 1. spectrum indicate that those signals have a broad fre- EnEx internal documents, Munich, 2012 quency range, typically between 20 kHz and 60 kHz, in contradiction to the emitted signal at 18 kHz. There- [4] B. Dachwald: Enceladus Explorer - A maneuverable fore those signals are transient signals which most likely subsurface probe for autonomous navigation through originate from cracks within the glacier. Furthermore it deep ice. 63th International Astronautical Con- gress, becomes clear that the noise level after averaging is still 2012 to high to achieve the desired detection. Both transient [5] R. Hoffmann: In-ice acoustic positioning system for signals and noise can be reduced by the use of frequency the Enceladus Explorer. ARENA 2012, AIP Confer- filters. The currently used filter is a Fourier filter with ence Proceedings, vol. 1535, pp. 200-203, 2013 a passband of 18 kHz to 26 kHz. The Fourier spectrum of the measured data as well as in red the frequency re- [6] D. Heinen et.al: Acoustic in-ice positioning in the sponse of the Fourier filter is shown in figure 3.2. Figure Enceladus Explorer project. Annals of Glaciology, 3.3 shows the Fourier filtered data. After filtering, the vol. 55 no. 68, 2014 expected signal at about 3 ms can be found by the FSM. [7] D. Eliseev et.al: Elektronikentwicklung f¨ur die The transient signals at the end of the waveform could akustische Ortung und Umfelderkundung einer Sonde be reduced but are still quite dominant. This shows that im Eis. DAGA 2014, Poster, 2014 the use of frequency filters is recommendable and is to be investigated further. The performed design study ver- [8] M. Meyer: Signalverarbeitung: Analoge und Digitale ifies that an infinite impulse response (IIR) filter [8] with Signale, Systeme und Filter. Vieweg+Teubner Verlag, a passband of 16 kHz to 23 kHz and a Chebyshev type 1 Wiesbaden, 2011 filter design [8] as shown in green in figure 3.2 gives the best results. Figure 3.4 shows the IIR filtered data. The

693